Many Kinds of Analytics, One Approach to Maximize Their Value

Working with clients it is clear that the interest in analytics has never been greater than it is today. Yet there is still confusion about what analytics means. We have web analytics to analyze logs and clickstream data to optimize websites; data mining and predictive analytics to analyze structured data to segment customers and to determine how likely something is to be true in the future; text analytics to analyze unstructured data and social media. All these approaches are useful if they are applied the right way and increasingly only if they are integrated.

Contents

Lack of integration is the biggest risk A focus on decisions maximizes analytic value Integration drives value and simplifies implementation

Lack of integration is the biggest risk

A lack of integration is the biggest risk as companies expand their use of analytics. Customers, suppliers and partners are increasingly multi-channel and any of them could be using the website. Structured and unstructured data could come from any of them and be about any of them. Successful companies are figuring out how to apply the right mix of techniques to the right problem. The best way to integrate these different kinds of analytics is to begin with the decision in mind. You must discover the decisions that matter to your business and understand what makes a good decision or a bad one. Only with this knowledge can you then drive back towards your data, determining the analytics that will help improve the decision and then identifying the data that is needed for those analytics.

A focus on decisions maximizes analytic value

A singular focus on decisions and on how analytics can improve them brings clarity to analytic efforts. It focuses different internal groups with different skills and tools on the same problem. It also makes clear what data must be integrated and what data quality issues must be resolved –the data needed for the decision must be integrated and good enough to support the analysis being proposed. No more absolute “all data must be integrated” projects that run on and on. No wasted time or money trying to hit some absolute level of data quality. Instead, Decision Management means results-focused, decision-centric projects that show tangible business benefits.

Integration drives value and simplifies implementation

Let’s discuss an example. Let’s say you have customer emails that someone wants to analyze with text analytics, CRM data and order history that is being used by data miners and an active product website that customers use to get manuals, product sheets and support. Web analytics, text analytics, predictive analytics are all in play but how to focus them? Think about your metrics and objectives. Is customer retention a focus or is it more important to increase the size of an initial order? Is expanding an existing customer footprint through cross-selling important or are you focused on acquiring as many new customers as possible? There’s no right answer for everyone but until you know what your answer is, you can’t make much forward progress.

If your company focused on customer retention you might want to improve the decisions you make about proactive retention actions. With this decision in mind it would be clear that predicting who’s at risk, what they are worth in future revenue and what kind of retention action would keep them loyal are all key analytics. Every group has something to offer in this context as the pattern of orders in the database, the kinds of information being viewed on the website by the customer and even the tone and number of support emails they send could all be highly predictive. A focus on the decision brings the need for integration and collaboration into focus.

Analytics of all kinds have much to offer but you need to focus on decisions to maximize their value.

[First published in Decision Management Solutions Newsletter April 5, 2011]